# Physically Partitioned KVCache Format for CPU--GPU Load Balancing in MoE Inference

arXiv: https://arxiv.org/abs/2609.14507 · HTML: https://mlsanity.com/paper/2609.14507/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Efficient MoE training and inference systems, one of 12 directions in Mixture-of-experts.
- Builds on 28 of its 49 references that are in the map.
- First posted on arXiv 13 Sep 2026.

## Placed in

- Mixture-of-experts / [Efficient MoE training and inference systems](https://mlsanity.com/moe/efficient-moe-training-and-inference-systems/)
- Efficient LLM inference and serving / [KV cache optimization](https://mlsanity.com/inference/kv-cache-optimization/)

## Builds on (most cited first, 28 of 28)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer](https://mlsanity.com/paper/1701.06538/) (2017)
- [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://mlsanity.com/paper/2101.03961/) (2021)
- [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023)
- [GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints](https://mlsanity.com/paper/2305.13245/) (2023)
- [DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model](https://mlsanity.com/paper/2405.04434/) (2024)
- [FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU](https://mlsanity.com/paper/2303.06865/) (2023)
- [FlashAttention-3: Fast and Accurate Attention with Asynchrony and Low-precision](https://mlsanity.com/paper/2407.08608/) (2024)
- [InfiniGen: Efficient Generative Inference of Large Language Models with Dynamic KV Cache Management](https://mlsanity.com/paper/2406.19707/) (2024)
- [PowerInfer: Fast Large Language Model Serving with a Consumer-grade GPU](https://mlsanity.com/paper/2312.12456/) (2023)
- [CacheGen: KV Cache Compression and Streaming for Fast Large Language Model Serving](https://mlsanity.com/paper/2310.07240/) (2023)
- [Mooncake: A KVCache-centric Disaggregated Architecture for LLM Serving](https://mlsanity.com/paper/2407.00079/) (2024)
- [LLM in a flash: Efficient Large Language Model Inference with Limited Memory](https://mlsanity.com/paper/2312.11514/) (2023)
- [Keyformer: KV Cache Reduction through Key Tokens Selection for Efficient Generative Inference](https://mlsanity.com/paper/2403.09054/) (2024)
- [PowerInfer-2: Fast Large Language Model Inference on a Smartphone](https://mlsanity.com/paper/2406.06282/) (2024)
- [vAttention: Dynamic Memory Management for Serving LLMs without PagedAttention](https://mlsanity.com/paper/2405.04437/) (2024)
- [MoE-Lightning: High-Throughput MoE Inference on Memory-constrained GPUs](https://mlsanity.com/paper/2411.11217/) (2024)
- [Fiddler: CPU-GPU Orchestration for Fast Inference of Mixture-of-Experts Models](https://mlsanity.com/paper/2402.07033/) (2024)
- [FastDecode: High-Throughput GPU-Efficient LLM Serving using Heterogeneous Pipelines](https://mlsanity.com/paper/2403.11421/) (2024)
- [Taming Latency-Memory Trade-Off in MoE-Based LLM Serving via Fine-Grained Expert Offloading](https://mlsanity.com/paper/2502.05370/) (2025)
- [HybriMoE: Hybrid CPU-GPU Scheduling and Cache Management for Efficient MoE Inference](https://mlsanity.com/paper/2504.05897/) (2025)
- [HeadInfer: Memory-Efficient LLM Inference by Head-wise Offloading](https://mlsanity.com/paper/2502.12574/) (2025)
- [LayerScope: Predictive Cross-Layer Scheduling for Efficient Multi-Batch MoE Inference on Legacy Servers](https://mlsanity.com/paper/2509.23638/) (2025)
- [Janus: Disaggregating Attention and Experts for Scalable MoE Inference](https://mlsanity.com/paper/2512.13525/) (2025)
- [KVDrive: A Holistic Multi-Tier KV Cache Management System for Long-Context LLM Inference](https://mlsanity.com/paper/2605.18071/) (2026)
- [An Efficient Hybrid Sparse Attention with CPU-GPU Parallelism for Long-Context Inference](https://mlsanity.com/paper/2605.07719/) (2026)
- [HybridGen: Efficient LLM Generative Inference via CPU-GPU Hybrid Computing](https://mlsanity.com/paper/2604.18529/) (2026)
- [ScoutAttention: Efficient KV Cache Offloading via Layer-Ahead CPU Pre-computation for LLM Inference](https://mlsanity.com/paper/2603.27138/) (2026)

## Abstract (arXiv)

Single-GPU long-context inference with Mixture-of-Experts (MoE) models requires spilling the key-value cache (KVCache) to CPU memory. The spilled KV serves two complementary purposes---transferring to the GPU for attention computation, or computing in-place on the CPU---which demand opposing physical states. The optimal split between them varies with workload, yet existing KVCache abstractions offer only storage semantics over a monolithic object of a single physical state, and cannot express dynamic load balancing. We propose InplaceKVCache, the first KVCache abstraction whose format fixes each byte's physical residency at write time, so that the CPU--GPU load balance can be adjusted without moving data after placement. It realizes this as a four-region layout along two dimensions---device affinity and access pattern---turning load balancing into pure scheduling. Built on this abstraction, WriteScope splits CPU--GPU shares along the sequence dimension, and a portable roofline performance model determines the optimal CPU share as sequence length evolves, with online feedback tracking CPU cost drift. On three MoE models (DeepSeek-V2-Lite, Qwen3-30B-A3B, Mixtral-8×7B) with a 32~GB VRAM budget, WriteScope supports end-to-end inference at the 1M-token aggregate scale. In the long-context regime (≥8K), it achieves geometric-mean speedups of 1.5×--2.5× on A100 and 1.4×--1.7× on V100 over four reproduced baselines, while vLLM, SGLang, and KTransformers fail even with a doubled KV budget. A DeepSeek-V4-Flash case study validates composition with native sparse attention.